The discussion centers on recent advancements in solving the Three Body Problem using deep artificial neural networks (DNNs). Two significant preprints are highlighted: one from Nature (https://www.nature.com/articles/s41586-019-1833-8) and another from arXiv (https://arxiv.org/abs/1910.07291). These studies demonstrate that DNNs can approximate the functions generated by numerical integrators, providing a constant-computational cost solution, although the term "solved" may be an overstatement.
PREREQUISITESResearchers in computational physics, machine learning practitioners, and anyone interested in the intersection of artificial intelligence and classical mechanics.
)"Solved" is maybe too strong a word for that paper... DNN's are trained to approximate the functions produced by numerical integrators. This leads to a constant-computational cost approximate solution. As far as I can tell anyways...gleem said: